Fix PR 23499

[Xinliang David Li]

Thanks. I have replied to the code review thread regarding this.

David


On Wed, May 27, 2015 at 11:06 AM, Eric Christopher <echristo at gmail.com>
wrote:

> I can. Brief summary: the counter data for a particular function needs to
> be in a comdat/elf section group with the function so that when a
> particular function is chosen by the linker it chooses the data for that
> function as well and then the other versions of the same function and their
> respective data are discarded.
>
> -eric
>
> On Wed, May 27, 2015 at 10:55 AM Xinliang David Li <xinliangli at gmail.com>
> wrote:
>
>> Diego, can you summarize what has been discussed?
>>
>> thanks,
>>
>> David
>>
>> On Wed, May 27, 2015 at 8:22 AM, Diego Novillo <dnovillo at google.com>
>> wrote:
>>
>>> Based on our IRC discussion, I think this will address the multiple
>>> copy problem on ELF systems.  By placing the counters in the same
>>> comdat section as its associated function, we get the linker to
>>> discard them together with the discarded function.
>>>
>>> I don't think we need to create another test case just to trigger this
>>> bug.  A small modification to the existing test case exposes it. I can
>>> create a new one, if you'd rather have a different test case for this,
>>> though.
>>>
>>> OK to commit?
>>>
>>>
>>> Thanks.  Diego.
>>>
>>> _______________________________________________
>>> llvm-commits mailing list
>>> llvm-commits at cs.uiuc.edu
>>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>>>
>>>
>> _______________________________________________
>> llvm-commits mailing list
>> llvm-commits at cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20150527/ab641b4c/attachment.html>